1. Purpose of the Appendix

This Appendix (hereinafter the "Appendix") defines the contractual framework applicable to the use of services provided by HumanLayer, including in particular:

• the allocation of roles and responsibilities between the parties,
• the governance and control rules applicable to Decisions rendered via HumanLayer,
• limitations and exclusions of liability,
• ethical principles and usage restrictions,
• traceability and auditability obligations,
• interaction with applicable regulatory and jurisdictional frameworks.

2. Definitions

For the purposes of this Appendix, the following terms have the meanings below:

• HumanLayer: the company providing a technical infrastructure for orchestrating human decisions.
• Client Organization: any contracting entity using HumanLayer services.
• AI Agent: any automated or semi-automated system initiating a Decision request via HumanLayer.
• Human Decision-Maker: any identified, qualified, authorized natural person rendering a human Decision via HumanLayer (referred to as “Sentinel” in non-contractual communications).
• Decision: any act of validation, refusal, approval, judgment, or authorization rendered by a Human Decision-Maker via HumanLayer.
• Decision SLA: indicative contractual time applicable to processing a Decision request.
• Decision Context: the set of information transmitted to the Human Decision-Maker to render a Decision.

3. Role and scope of HumanLayer

3.1 Exclusive technical orchestration role

HumanLayer acts exclusively as a technical provider of human decision orchestration.

As such, HumanLayer:

• does not make any Decisions on the merits,
• does not exercise decision authority,
• does not modify or alter the content of a Decision,
• does not provide any final recommendation.

3.2 No autonomous judgment or influence

HumanLayer:

• does not analyze or evaluate the Decision Context,
• does not interpret transmitted information,
• does not exert any direct or indirect influence on the reasoning, assessment, or choice of the Human Decision-Maker.

3.3 Anti-requalification clause

No provision of this Appendix shall be interpreted as granting HumanLayer a role of Human Decision-Maker, co-Human Decision-Maker, agent, representative, or decision auxiliary for the Client Organization or the Human Decision-Maker.

4. Responsibilities

4.1 Human Decision-Maker responsibility

Each Decision rendered via HumanLayer is taken by a Human Decision-Maker:

• identified,
• qualified,
• authorized under applicable professional and regulatory rules,

who assumes full professional, civil, criminal, and regulatory liability within the limits provided by law.

4.2 Client Organization responsibility

The Client Organization remains solely responsible for:

• defining its decision policies,
• the use and execution of Decisions rendered,
• integrating Decisions into its internal processes,
• compliance with laws and regulations applicable to its activities,
• supervision of the AI Agents it deploys.

4.3 HumanLayer responsibility

HumanLayer's liability is strictly limited to:

• providing the technical orchestration service,
• compliance with contractually defined availability and security commitments,
• faithful traceability of Decisions rendered.

HumanLayer assumes no responsibility for:

• content,
• appropriateness,
• legality,
• or the consequences of Decisions rendered.

5. Governance and control

5.1 Decision policies

HumanLayer provides technical means allowing the Client Organization to apply its own decision policies.

HumanLayer does not define, interpret, or evaluate these policies.

5.2 Human Decision-Maker verification

The verification mechanisms implemented by HumanLayer are intended solely to check the declarative existence of qualifications or authorizations communicated by Human Decision-Makers.

These verifications do not constitute in any way:

• a certification,
• a guarantee,
• or a validation of competence.

5.3 Deactivation and control

The Client Organization may at any time:

• restrict certain categories of Decisions,
• suspend API calls,
• disable HumanLayer services entirely.

6. Risk and incident management

6.1 Decision error

A Decision error does not engage HumanLayer's liability as long as the Decision was rendered by an identified Human Decision-Maker based on traceable Decision Context.

6.2 Erroneous or incomplete context

HumanLayer keeps a faithful trace of the Decision Context as transmitted by the AI Agent.

It is up to the Human Decision-Maker to accept, refuse, or request additional information.

6.3 Decision SLAs

Decision SLAs constitute obligations of means, not result.

7. Ethics et exclusions d'usage

7.1 Prohibited uses

Use of HumanLayer is strictly prohibited for any Decision aiming in particular to:

• circumvent applicable law or regulation,
• violate fundamental rights,
• conceal or artificially transfer human accountability,
• validate after-the-fact a decision already executed,
• exert coercion or manipulation on a Human Decision-Maker.

7.2 Ethical right of refusal

HumanLayer reserves the right to refuse, suspend, or terminate any use that does not comply with the principles set out in this Appendix.

8. Traceability and audit

HumanLayer logs the following elements:

• Human Decision-Maker identity,
• Decision Context transmitted,
• Decision rendered,
• justification provided,
• timestamp.

These elements may be made available to the Client Organization for audit purposes, under the applicable contractual conditions.

9. Regulatory framework and jurisdictions

9.1 Human oversight

HumanLayer is designed to enable effective human oversight in accordance with applicable regulations for automated systems and AI.

9.2 Jurisdictions

Each Decision is rendered by a Human Decision-Maker declaring they are authorized to act in the applicable jurisdiction, under the exclusive responsibility of the Human Decision-Maker and the Client Organization.

10. Mutual indemnification

10.1 Indemnification by the Client Organization

The Client Organization agrees to indemnify and hold HumanLayer harmless against any claim, action, or judgment resulting from:

• the use of Decisions rendered,
• a breach by the Client Organization of its legal obligations,
• use of the services in violation of this Appendix.

10.2 Indemnification by HumanLayer

HumanLayer agrees to indemnify the Client Organization against any final judgment resulting exclusively from:

• a proven breach by HumanLayer of its contractual obligations,
• directly attributable to the provision of the orchestration service.

11. Non-delegation of responsibility

Human accountability remains full, complete, and non-transferable.

12. Final provisions

This Appendix is binding on the parties and applies to any use of HumanLayer services.